\begin{table}[h!]
    \centering
    \caption{Robustness to Lot Area Threshold}
    \label{tab:lot_area_thresh}
    \begin{tabular*}{\textwidth}{@{\extracolsep{\fill}}lccc@{}}
\toprule & (1) & (2) & (3) \\ \addlinespace 
                     & \multicolumn{3}{c}{Lot Area Threshold} \\ \cmidrule{2-4}
                     & 20\%                       & 10\%                       & 30\%                       \\  \midrule \addlinespace 
No Controls          &  $$logit1_20_starbeta$$                         &        $$logit1_10_starbeta$$                    &       $$logit1_30_starbeta$$                     \\
                     &  ($$logit1_20_se$$)                         &         ($$logit1_10_se$$)                    &           ($$logit1_30_se$$)                  \\ \addlinespace 
Lot Controls         &   $$logit2_20_starbeta$$                         &        $$logit2_10_starbeta$$                    &       $$logit2_30_starbeta$$                     \\
                     &  ($$logit2_20_se$$)                         &         ($$logit2_10_se$$)                    &           ($$logit2_30_se$$)                  \\ \addlinespace 
Block Controls       &     $$logit3_20_starbeta$$                         &        $$logit3_10_starbeta$$                    &       $$logit3_30_starbeta$$                     \\
                     &  ($$logit3_20_se$$)                         &         ($$logit3_10_se$$)                    &           ($$logit3_30_se$$)                  \\ \addlinespace 
Lot \& Block Controls &       $$logit4_20_starbeta$$                         &        $$logit4_10_starbeta$$                    &       $$logit4_30_starbeta$$                     \\
                     &  ($$logit4_20_se$$)                         &         ($$logit4_10_se$$)                    &           ($$logit4_30_se$$)                  \\ \addlinespace \bottomrule
\end{tabular*}
            \begin{spacing}{1}
\begin{tablenotes}
      \item \footnotesize \textit{Notes:} This table examines the robustness of the GMM logit coefficient estimates for the effect of the 421-a tax incentive. The specification estimated is always Equation \ref{eq:gmm}, changing the construction of the simulated instrument. Column 1 reports baseline estimates, in which we use the simulated instrument where comparable lots must be within 20 percent of the lot area as the actual lot. Column 2 tightens this comparison to lots within 10 percent of the actual lot area. Column 3 loosens the comparison to 30 percent. Each row presents a specification within a given set of controls. All specifications include fixed effects for the borough and year of initial permit issuance.  Standard errors are clustered at the neighborhood level. $\sym{*} = p < 0.10$, $\sym{**} = p < 0.05$, $\sym{***} = p < 0.01$.
      \end{tablenotes}
       \end{spacing}
\end{table}